Re: [PATCH v2 2/3] MdePkg/BaseMemoryLib*: add missing ASSERT()s

[Ard Biesheuvel <ard.biesheuvel@linaro.org>]

On 20 September 2016 at 13:16, Ard Biesheuvel <ard.biesheuvel@linaro.org> wrote:
> On 20 September 2016 at 13:02, Wu, Hao A <hao.a.wu@intel.com> wrote:
>>> -----Original Message-----
>>> From: edk2-devel [mailto:edk2-devel-bounces@lists.01.org] On Behalf Of Ard
>>> Biesheuvel
>>> Sent: Tuesday, September 20, 2016 4:38 PM
>>> To: Wu, Hao A
>>> Cc: leif.lindholm@linaro.org; edk2-devel@lists.01.org;
>>> vishalo@qti.qualcomm.com; Gao, Liming
>>> Subject: Re: [edk2] [PATCH v2 2/3] MdePkg/BaseMemoryLib*: add missing
>>> ASSERT()s
>>>
>>> On 20 September 2016 at 03:00, Wu, Hao A <hao.a.wu@intel.com> wrote:
>>> > Hi Ard,
>>> >
>>> > The NULL checks for the input Guids in APIs CopyGuid(), CompareGuid() and
>>> > IsZeroGuid() are implicitly done within calls to BaseLib APIs
>>> > ReadUnaligned64() and WriteUnaligned64().
>>> >
>>> > So I think the functions behavior matches with their comments. What do you
>>> > think?
>>> >
>>>
>>> I disagree. ReadUnaligned64 and WriteUnaligned64 could theoretically
>>> be implemented by a version of BaseLib that does not contain such
>>> ASSERT()s
>>
>> The comments for APIs ReadUnaligned64 and WriteUnaligned64 in BaseLib
>> mention the ASSERT() for inputting a NULL buffer.
>>
>> I think instances of BaseLib should follow the comments.
>>
>
> I agree with this. But that does not justify omitting them here.

... but actually, it makes no sense to argue about this, since the
user visible outcome is the same. I will withdraw the patch.

Thanks,
Ard.